Importance Septic arthritis of the native knee joint is the most common bacterial joint infection. The management involves prompt surgical debridement and joint irrigation by arthroscopy or arthrotomy. This is the first systematic review and meta-analysis to compare arthroscopic debridement with arthrotomy for septic arthritis of native knee joint.
Objective The purpose of this systematic review and meta-analysis is to compare re-operation rates, length of inpatient hospital stay (LOS) and functional outcome between arthroscopy and arthrotomy in the treatment of acute septic arthritis of the native knee joint.
Evidence review This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. PubMed, Embase and Cochrane Central Register of Controlled Trials databases were searched from database inception to 31 May 2019. All original studies that compared re-operation rates and LOS between arthroscopy and arthrotomy for septic arthritis of knee were included. The research question and eligibility criteria were established a priori. Pertinent data were extracted and random-effects model was used to pool the data where possible.
Findings A total of seven studies with 1089 knees were included, of which 723 underwent arthroscopic surgery and 366 knees underwent arthrotomy. The relative risk of re-operation was significantly lower in the arthroscopy group with a pooled relative risk of 0.69 (95% CI 0.56 to 0.86; p=0.0006). All studies reported shorter LOS and one study reported better functional outcomes in the arthroscopy group as compared with arthrotomy. However, the data could not be quantitatively synthesised due to variation in reporting among the studies included.
Conclusions and relevance Based on the available evidence, we conclude that arthroscopy for the treatment of septic arthritis of the knee results in a lower re-operation rate than arthrotomy. It cannot be concluded whether arthroscopic treatment results in shorter LOS or better functional outcome as compared with arthrotomy.
Level of evidence IV
- septic arthritis
Statistics from Altmetric.com
What is already known
Acute septic arthritis of knee requires prompt surgical debridement and joint washout.
It is unclear if there is any difference in the re-operation rates or length of hospital stay, when this procedure is done arthroscopically or via arthrotomy.
What are the new findings
Arthroscopic debridement and joint lavage for septic arthritis of knee has lower re-operation rates than standard arthrotomy.
By virtue of being less invasive, arthroscopic treatment has a trend towards lesser length of stay as compared with arthrotomy.
There is no meta-analysis or large adequately powered randomised controlled trial to date demonstrating any difference between the two procedures for septic arthritis of native knee.
Septic arthritis is considered to be a surgical emergency with delayed diagnosis and treatment being associated with significant morbidity and mortality.1–3 The incidence of bacterial joint infection is around 4–10 per 100 000 patients per year. The knee joint is the most affected joint and is involved in about 50% of the cases.4–8
The management of septic arthritis involves prompt surgical intervention, joint irrigation, removal of purulent material and intravenous antibiotics.9 Surgical debridement and joint irrigation can be performed either arthroscopically or via arthrotomy. Several studies have compared arthrotomy and arthroscopic treatment for septic arthritis of knee.10–17 To the best of our knowledge, there has not been any systematic review or meta-analysis in the English literature comparing arthrotomy and arthroscopy in the treatment of septic arthritis of the knee.
The purpose of this systematic review and meta-analysis is to compare re-operation rates and length of inpatient hospital stay (LOS) between arthroscopy and arthrotomy in the treatment of acute septic arthritis of the knee.
Material and methods
Protocol and registration
The systematic review and meta-analysis was performed using a predetermined protocol in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.18 The study protocol was registered with PROSPERO (CRD42018086676).
Search strategy and study selection
The online databases PubMed, Embase and Cochrane Central Register of Controlled Trials were searched from database inception until 31 May 2019. Additional search of electronic publications ahead of print was done in PubMed using the syntax published earlier.19
We searched for studies that compared treatment of septic arthritis of knee with arthrotomy and arthroscopy, using a combined text and MeSH search with the following terms: ‘septic arthritis’, knee’ and ‘arthroscopy’.
The articles were selected in two stages (figure 1). First, the abstracts identified by the above searches were analysed and the list was screened using the inclusion and exclusion criteria. Second, the full texts of this shortlisted list were downloaded and assessed for eligibility. The reference lists of the publications were then hand-searched for additional relevant studies. This process was repeated twice independently.
All original studies were included if (1) the re-operation rates due to failure of index procedure and/or LOS were compared between arthrotomy and arthroscopy, (2) the re-operation rates and/or LOS and/or functional outcomes could be individually retrieved for each group and (3) the studies were available in the English language. The articles were included or excluded as demonstrated by the flow chart in figure 1.
Two reviewers used a standardised form to extract data. Data regarding the authors, year of publication, location of study, study design, level of evidence (American Academy of Orthopedic Surgeons grades of evidence), sample size, number of re-operations, LOS and functional outcomes were extracted.20 Re-operations due to any cause other than infection were excluded from the analysis. In publications where the definitions were not clear or the data were inadequate, we tried to contact the authors of the respective studies to request specific information needed. If we were unable to contact the authors of these publications to get the required data, the study was excluded from the analysis.
The methodological quality of the included studies was assessed by the same two reviewers. The Cochrane collaboration’s risk of bias and Risk Of Bias In Non-randomised Studies-of Interventions tool was used for randomised controlled trials and retrospective cohort studies, respectively.21 22 Additionally, the Methodological Index for Non-randomised Studies (MINORS)) instrument was used to assess the methodological quality of comparative and non-comparative, non-randomised surgical studies.23 The MINORS checklist assigns a maximum score of 24 for comparative studies. Any disagreements were discussed between the reviewers until consensus was reached.
Meta-analysis was performed to compare outcomes between arthroscopy and arthrotomy. The random-effects model was used to obtain pooled estimates for each outcome.24 This assumes that the studies represented a random sample from the larger population of such studies, with each study having its own underlying effect size. Under this random-effects model, it is assumed that there is a mean population effect size about which the study-specific effect varies. As the random-effects model properly takes into account the interstudy heterogeneity such as differences in study design and definitions of outcomes, it thus provides a more conservative evaluation of the significance of the association than one based on fixed effects.25
The relative risk (RR) is reported with the 95% CI. Tests of heterogeneity were conducted with the Q statistic that is distributed as a χ2 variate under the assumption of homogeneity of effect sizes. The extent of between-study heterogeneity was assessed with the I2 statistic.26 27
Publication bias was investigated graphically using funnel plots.28 29 Funnel plots were created by plotting the ratio of RR against reciprocal of its SE. Sensitivity analysis was conducted to assess the potential impact of modelling assumption on the study findings.
The level of significance was set at p<0.05. All statistical analyses were carried out using RevMan V.5.3.30
A total of seven studies with 1089 knees were included, of which 723 underwent arthroscopy and 366 knees underwent arthrotomy. The study baseline characteristics are summarised in table 1. Two studies were from the USA, three from Europe, one from Australia and one from Brazil. Six studies were retrospective cohort studies (level 3) and one was a prospective randomised controlled trial (level 1). None of the studies had a critical risk of bias (table 2).10–13 15–17 Of the studies, three cohort studies were assessed to have a high or serious risk of bias owing to confounding and possible bias due to missing data in these studies. Only one study was a double-blinded randomised controlled trial and had an overall moderate risk of bias due to missing data from loss to follow-up.16 The remaining three cohort studies were assessed to have moderate risk of bias due to confounding.10 11 15 The average MINORS score for the cohort studies included in the meta-analysis was 15.6 out of the maximum possible score of 24 (table 1).
The number of re-operations in patients undergoing arthrotomy and arthroscopy was reported in all seven included studies with 1089 knees.10–13 15–17 Overall, there were 124 re-operations out of 723 (17%) in the arthroscopy group and 81 re-operations out of 366 (22%) in the arthrotomy group. The pooled RR of re-operation was 0.69 (95% CI 0.56 to 0.86; p=0.0006) favouring the arthroscopy group. The studies were homogenous (I²=0%; τ²=0.00; χ²=5.59, df=6; p=0.47). Subgroup analysis of the six non-randomised cohort studies and one randomised controlled trial are shown separately in the forest plot (figure 2). The directions of the randomised controlled trial and the cohort subgroups were similar. Hence, the subgroups were combined to give a pooled ratio of RR.
Length of in-hospital stay
LOS was reported in three studies with 282 knees with septic arthritis.11 15 17 One study reported the information using mean and range, whereas two studies reported the information using median and IQRs. Usually, median is reported by the authors when the data are not normally distributed. Hence, this outcome could not be pooled and synthesised quantitatively. However, the arthroscopy group showed a shorter LOS in all the papers available. These results are presented in table 3.
Postoperative functional outcomes were reported in three out of eight studies.10 16 17 One studies reported Lysholm scores, one study reported Larson score and one study reported Bussiere and Beaufils functional scale. Due to the different scales of reporting functional outcomes in the different studies, the functional outcomes could not be pooled and synthesised quantitatively. The summary is present in table 4.
There was no publication bias as observed from the funnel plot (figure 3). Additionally, we performed the Egger’s test which also suggested no publication bias.
To verify the robustness of our findings, meta-analysis was repeated by application of fixed-effect model to our data. The pooled RR for re-operations was 0.68 (fixed effect, 95% CI 0.54 to 0.86), which is similar to that of our random-effects model.
The most significant finding of this meta-analysis was that the risk of re-operation following arthroscopy was significantly lower as compared with arthrotomy (RR=0.69; 95% CI 0.56% to 0.86%) in the treatment of acute septic arthritis of the knee. All studies which reported the LOS also showed a shorter LOS in the arthroscopy group as compared with arthrotomy group (table 3). However, we were unable to pool the data for LOS due to variation in reporting among the studies included. To the best of our knowledge, this is the first systematic review and meta-analysis to compare arthroscopy and arthrotomy in the treatment of septic arthritis of knee.
Arthroscopy, as compared with arthrotomy, in the treatment of septic arthritis of the knee has advantages of being minimally invasive, thereby causing less pain, enabling a quicker postoperative recovery and resulting in a shorter LOS.31 32 However, there are concerns of its efficacy in controlling knee infection due to difficult visualisation, inadequate drainage and incomplete synovectomy.10
The study by Johns et al was not included in the meta-analysis as it was based on paediatric population and the other studies included adults. However, the results suggests that the better results of the arthroscopy in the adult population hold true in the paediatric population.14 The study by Peres et al was the only randomised controlled trial comparing arthrotomy and arthroscopy for septic arthritis of knee.16 It included 21 cases in all (11 patients who underwent arthroscopy and 10 patients who underwent arthrotomy). As all patients in the study were free of infection at final follow-up of 15 months, the study reported no difference in the effectiveness of treatment between arthroscopy and arthrotomy in the treatment of septic arthritis of knee. However, in the arthrotomy group, they reported two re-operations due to recurrence of infection and significantly higher pain in the arthrotomy group at 7 and 14 days postoperatively. Pain score at 21 days postoperatively were not different among the two groups.
This is similar to five out of six other retrospective cohort studies, which also reported lower re-operation rates in the arthroscopy group as compared with the arthrotomy group.11–15 17 In contrast, Balabaud et al 10 reported higher re-operation rates in the arthroscopy group as compared with arthrotomy. However, the failure of treatment was likely correlated to the time delay prior to surgery and not related to the surgical procedure (12 days for cases treated successfully vs 23 days for treatment failure). Böhler et al 11 also reported that cases in which treatment failed had symptoms for 7 days as compared with 4 days for treatment success. Wirtz et al 17 similarly suggested better outcomes when treatment was undertaken early (within 5 days of onset). They suggested arthrotomy for cases where there was delay in treatment of >5 days or extensive synovial and capsular involvement. Faour et al and Bovonratwet et al,13 33 both reported data from the American College of Surgeons National Surgical Quality Improvement Programme database. Faour et al 13 reported data between 2011 and 2015, whereas Bovonratwet et al 33 reported data between 2005 and 2014. As there was overlapping data in the two studies, Faour et al 13 was considered for analysis as the study had double the cohort size. They reported lower re-operation rates and early discharges in patients treated with arthroscopy as compared with arthrotomy.
The possible reasons for the greater efficacy of arthroscopy as compared with arthrotomy in the treatment of septic arthritis of the knee can only be hypothesised. It is established that polymorphonuclear leucocytes demonstrate reduced chemotaxis, respiratory activity and bactericidal capacity at reduced pH.34 A hypothesis could be that the longer incision and local tissue injury during arthrotomy may lead to a reduced local pH due to postoperative inflammatory response.35 This in turn may cause a reduction in local immunity and higher failure rates. Another possible reason is that as the joint is relatively closed during arthroscopy, the joint space may be more thoroughly irrigated, as compared with arthrotomy where some pockets may inadvertently escape irrigation.
Apart from re-operation rates, two studies reported preoperative and postoperative range of motion in the two groups.11 17 They were not significantly different in the two groups.11 However, these measurements were taken at different time points after the procedure in the two studies. Peres et al 16 also reported preoperative and postoperative knee functional scores (Lysholm scores), return to activities of daily living, pain scores (21 days postoperatively) and range of motion was not significantly different in the two groups. None of the studies reported postoperative knee radiographs for radiological assessment and progression of osteoarthritic changes.
While this study represents the first systematic review and meta-analysis comparing the rates of re-operation and LOS between arthroscopy and arthrotomy in the treatment of septic arthritis of the knee, this study faces several limitations. First, there may be clinical and methodological heterogeneity between the included studies. The infecting micro-organism can have an effect on the outcome of septic arthritis. For example, methicillin-resistant S taphylococcus aureus septic arthritis has been shown to be associated with poor outcomes as compared with other organisms.36 Two studies in our review reported the organism isolated in each group.14 15 However, outcomes (re-operations) based on the micro-organism isolated were not reported in any study under review. The difference in the microbiological profile of the patient groups in the various studies may therefore be a source of clinical heterogeneity. Furthermore, surgeons treating septic arthritis of the knee may have varying thresholds for re-operation based on clinical judgement. This may be a source of methodological heterogeneity. However, the outcomes were homogenous across the different studies, and there was no publication bias identified as per the forest plots and Egger’s test.
Second, there could be potential selection bias in the studies included in the meta-analysis. As there was no randomisation in five out of six included studies, it is possible that clinically less severe cases of septic knee arthritis were selected for arthroscopy resulting in better outcomes in that group. Despite this, the only randomised controlled trial published on this topic also similarly showed that arthroscopy had a lower risk of re-operation as compared with arthrotomy in the treatment of septic arthritis. Also, the absence of statistical heterogeneity suggests that the outcomes could have been unaffected by the potential selection bias.
Lastly, the level of evidence of this review is limited by the lack of high-quality trials available in the literature. There was a total of only six retrospective cohort studies (level 3) and one randomised controlled trial (level 1) that were available in the existing literature. This systematic review and meta-analysis then highlights that the knowledge on this topic is currently in its infancy, and further work can be performed to confirm or refute the current knowledge in the field.
Based on the available evidence, we conclude that arthroscopy for the treatment of septic arthritis of the knee results in a lower re-operation rate than arthrotomy. It cannot be concluded whether arthroscopic treatment results in shorter LOS or better functional outcome as compared with arthrotomy.
Contributors All of the authors have contributed to the concept, design and preparation of the manuscript. TP and KLW contributed to data abstraction, review of literature and writing of the manuscript. TS and GL contributed to the writing and review of the literature. NV and LK contributed to the review of the literature, manuscript and editing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.